Drug models of schizophrenia

The extracted text presents this article as a selective narrative review rather than a systematic review or meta-analysis. The authors do not provide an explicit methods section in the extracted text describing literature search strategies, inclusion/exclusion criteria, databases searched, or formal risk-of-bias assessment. Consequently, the specific criteria used to select studies and the comprehensiveness of the search are not clearly reported in the provided material. Interpretation throughout the paper is therefore based on synthesis of findings from human experimental studies, clinical observations, imaging and post-mortem studies, and animal models, with cross-references between neurochemical evidence and behavioural effects. Where animal data are discussed, the authors note advantages (e.g. possibility of chronic or perinatal dosing) and limitations (translation of uniquely human symptoms and biomarkers). No quantitative meta-analytic methods or statistical pooling are described in the extracted text.

Dopaminergic models: The authors reiterate the long-standing relationship between antipsychotic efficacy and D2 receptor affinity and review evidence for abnormal dopaminergic function in schizophrenia, including increased dopamine release after amphetamine, elevated synaptic dopamine and increased [18F]DOPA uptake. Psychostimulants such as amphetamine and cocaine raise synaptic dopamine and can precipitate or exacerbate psychosis; large single oral doses of amphetamine produced acute psychosis in early volunteer studies, while lower doses tend to produce paranoid symptoms only after repeated dosing in vulnerable individuals. In animals, dopaminergic stimulants induce stereotyped behaviours and impairments in prepulse inhibition (PPI), with chronic dosing producing sensitisation. The dopaminergic model is characterised as a good model of paranoid/positive symptoms but limited for cognitive and negative domains; reliance on dopaminergic models may bias drug development toward further dopamine-targeting agents. Glutamatergic (NMDA receptor) models: The NMDA receptor hypofunction hypothesis is presented as an alternative or complementary model. Genetic and post-mortem data provide some support for glutamatergic involvement, and uncompetitive NMDA antagonists PCP and ketamine induce increases in gamma oscillatory power, functional connectivity changes, and rises in prefrontal glutamate or glutamine measured by microdialysis and 1H-MRS. Mechanistic accounts include reduced GABAergic interneuron function and consequent disinhibition of pyramidal cells, though the precise cellular targets of ketamine are debated. Ketamine produces positive, negative and cognitive symptoms in healthy volunteers and can exacerbate symptoms in patients; binding to NMDA receptors correlates with negative symptoms while downstream glutamatergic changes correlate with positive symptoms. Chronic exposure to PCP/ketamine in humans and animals can produce persistent schizophrenia-like symptoms and neurobiological changes — for example, reduced prefrontal grey matter, decreased thalamic NAA, upregulation of frontal D1 receptors, and cortical neurodegenerative changes in rodents — which are partially attenuated but not fully reversed by antipsychotics. Serotonergic models: Classical serotonergic hallucinogens (LSD, psilocybin) are 5-HT2A receptor agonists and induce agitation, visual hallucinations and disruptions in PPI and brain network integration, with downstream increases in glutamate. These effects resemble aspects of prodromal or first-episode psychosis more than established chronic schizophrenia, in part because hallucinations are more commonly visual. Tolerance develops to psychedelics, limiting their capacity to model chronic changes, though chronic low-dose administration in rodents has produced persistent behavioural syndromes resembling several symptom domains that respond to antipsychotics. Endocannabinoid models: Acute THC administration in healthy people can induce positive and negative symptoms and cognitive impairments and exacerbate illness in patients, with associated changes in neurophysiological measures (P50 suppression, mismatch negativity, P300) and perceptual measures such as altered binocular depth perception. Animal studies show dose-dependent working memory deficits and PPI impairment, with greater vulnerability during adolescent exposure. Cannabidiol (CBD) is discussed as having anxiolytic and putative antipsychotic properties; a recent double-blind placebo-controlled trial in early-stage schizophrenia reported efficacy of CBD versus psychotic symptoms similar in magnitude to existing antipsychotics but with fewer side effects. CBD also attenuated some ketamine-induced effects, suggesting interactions between the endocannabinoid and glutamatergic systems. GABAergic models: Post-mortem evidence indicates altered GABAergic neurons and receptor subunit expression in schizophrenia, and GABAergic signalling interacts closely with glutamate and dopamine. GABA-A antagonists disrupt PPI when infused into rodent medial prefrontal cortex, an effect reversible with D2-blocking antipsychotics. Animal models (including methylazoxymethanol acetate) suggest that hippocampal GABAergic deficits may drive increased striatal dopamine activity; positive allosteric modulation of alpha-5 GABA-A receptors reduced hyperactivity in such models. Human pharmacological manipulations yield mixed effects: iomazenil (a benzodiazepine receptor antagonist) worsened psychotic symptoms in patients, while some GABA-A agonists can worsen schizophrenia possibly via presynaptic actions. Overall, clinical results for GABA-targeting compounds have been disappointing so far. Cholinergic models: Nicotinic and muscarinic systems are implicated. High tobacco use in schizophrenia has prompted hypotheses of self-medication; post-mortem reductions in alpha-4 and alpha-7 nicotinic receptors have been reported and alpha-7 agonists have shown early promise for cognitive and negative symptoms. Varenicline did not substantially worsen psychosis in patients. Muscarinic antagonists (atropine, scopolamine) produce delirium, visual hallucinations and cognitive deficits, and reduced muscarinic receptor availability has been observed in unmedicated first-episode patients. Clozapine’s muscarinic occupancy and xanomeline’s M1/M4 partial agonism — the latter showing efficacy versus positive and negative symptoms and cognition in early trials — are noted, though xanomeline’s cholinergic side effects limited tolerability. Kappa-opioid models: Salvia divinorum, a kappa opioid agonist, produces powerful dissociative and hallucinatory experiences often involving encounters with other realms or entities; such phenomenology differs from typical schizophrenic psychosis. Small-scale human data and case reports indicate possible links between kappa systems and symptomatology, and altered dynorphin levels and receptor distribution have been suggested in schizophrenia, but the evidence base is limited and requires further study. Convergence via prefrontal glutamate: Multiple pharmacological manipulations — acute NMDA blockade, 5-HT2A agonism, THC and amphetamine — have been reported to increase prefrontal glutamate (measured by microdialysis or estimated by 1H-MRS). Early psychosis has been associated with increased prefrontal glutamine and elevated glutamate measures have been linked to failure to remit on dopaminergic antipsychotics. The authors note conflicting 1H-MRS findings and discuss the difficulty of distinguishing neurotransmitter pools from metabolic pools of glutamate and glutamine; 13C MRS is identified as a more direct but costly method.

The authors interpret the assembled evidence by mapping strengths and limitations of each pharmacological model against symptom domains of schizophrenia. Dopaminergic stimulants are judged to model paranoid positive symptoms well but to fall short in reproducing negative symptoms and cognitive deficits. In contrast, NMDA receptor antagonists (PCP, ketamine) and THC are presented as producing a broader spectrum of schizophrenia-like features, encompassing positive, negative and frontal cognitive symptoms, and as inducing persistent neurobiological changes with chronic exposure that resemble aspects of the disorder. Steeds and colleagues emphasise that no single pharmacological model can capture the chronic, neurodevelopmental and episodic nature of schizophrenia, and they caution that some drug-induced states differ phenomenologically from schizophrenia (for example, visual rather than auditory hallucinations, or the experience of euphoria). The authors also warn that attenuation of drug-induced effects by an antipsychotic in experimental models may reflect pharmacological interactions irrelevant to schizophrenia pathogenesis, and that animal findings require careful cross-validation with human data. Key uncertainties and limitations are acknowledged: the extracted text notes conflicting findings in 1H-MRS studies of medial prefrontal metabolites, incomplete understanding of cellular targets for drugs like ketamine, and small-scale or sparse data for systems such as kappa opioids. Tolerance and the unpredictability of psychedelic experiences are raised as constraints on serotonergic models, while chronicity and ethical/practical issues limit the use of some stimulants in experimental paradigms. The authors also report that clinical trials for some receptor-targeted agents (for example 5-HT2A antagonists) have failed to demonstrate efficacy for positive symptoms in later-phase trials despite mechanistic rationale. In terms of implications, the review suggests that NMDA antagonist and endocannabinoid models may be particularly useful for identifying therapies targeting cognitive and negative symptoms, and that further exploration of GABAergic and cholinergic targets remains promising despite mixed clinical results to date. The authors advocate development of translational animal models informed by human studies and the use of objective measurements to predict therapeutic efficacy and to dissect abnormalities in specific brain circuitry. A central recommendation is to study interactions between neurotransmitter systems — particularly how various manipulations converge on prefrontal glutamate — as an important step toward a coherent hypothesis of schizophrenia pathogenesis.

Steeds and colleagues conclude that different pharmacological models capture different facets of schizophrenia: dopaminergic stimulants model paranoid positive symptoms effectively, but NMDA receptor antagonists and THC offer more complete models that include negative and cognitive domains and can induce persistent neurobiological changes in chronic exposure. Other systems — serotonergic, GABAergic, cholinergic and kappa opioids — may model aspects of prodromal or specific symptom clusters and could yield novel therapeutic targets. While pharmacological models cannot fully replicate the complexity of schizophrenia, they remain valuable for probing symptom-specific neurobiology and for guiding drug development. The authors call for integrated study of interacting neurotransmitter systems and for translational work that links human and animal findings to accelerate the discovery of improved treatments.